Application Design Question Lage Data Items

I’m working on an application that needs to be disconnected at times.

The SaveCacheState() and RestoreCacheSate() looks like it will work well for most of the application architecture.

The question I have is centered around (for example) the use of large image files for instance.

Let’s say the application (WinForm and or Silverlight) would contain thousands of pictures. I would not want to make a wire trip to hydrate them on the client; I want to store them locally. My method prior to DevForce was to have an instance of SQLExpress on the client and store them there as blobs.  I would then use Merge Replication to keep the data in sync when connected.

So the question I have is could I and or should I store these bits in a Cache file?

Is there a limitation on the size and what would the performance penalty be for “finding” the correct picture to display? When I would want to show a picture how is it retrieved from the cache it the entire cache need to be expanded or is the BOS smart enough to just extract the one I need from the file.

In the Hello DevForce App it shows a picture for each Employee if we had a DB that had 20,000 employees in it and it needed to be disconnected what could be an approach to achieving this.

Any design pattern around the architecture would be appreciated. Preferably one that would work for both a Win Client and a OOB SilverLight App.

What I’m thinking I want to do to limit network bandwidth as well as make full disconnected application availability to “know” data is have a local instance of a database say using sqlexpress. The program logic flow I’m thinking about is this.

When the UI (end user) requests a “picture” if first looks for it in the cache if it’s not there is looks for it in the local database if it’s not there it then tries to get it for the remote database.  When it is found it will place it in the local database.  Once it’s in the Local DB there should never be wire trip to get that image. Each image by design is static, if the image changes a new one will be added (old one keep for historic reasons)

So using DevForce, the BOS, Cache, and EntityManager etc I’m trying to think of a way to implement this logic. And if this sounds like it will work ok.

In a nut shell I only want to use the SaveCacheState() and RestoreCacheState() when:

1)      Getting data to work offline

2)      Saving changes after working offline

3)      Saving changes back to the Master Remote DB after coming back online

There is a brief section in the Doc under Business Object Persistence – Advanced called “Access Both Local and Remote Data Sources in the Same N-tier Application” This is what I want to do.

If I understand it correctly I will need 2 EntityManager all sounds good. I think I get it.

Now the design question.  If I have a table called ProductImages that hold 300,000 + images on the remote server, and let’s say I have already loaded all that data to my local DB, when I’m disconnected would I use the Cache or my localDB. If I use the Local_EntityManager the stuff will be saved to LocalDB great but how would I get that data back to the remote DB when I come back online. I’m trying to avoid using Merge Replication.

So any high level design recommendations would be great. I would like to see a reference application of this design pattern.

 I would be willing to help provide one as I think this would be a common implementation and a reference app from DevForce would be cool.

Greg,

                Thanks for taking the time to answer.

>> 

Given your relatively complex algorithm for retrieval of a particular image for immediate use (first look in cache; then in the local db; then in the remote…

>> 

Well this is what I thought I would want to do, maybe not. What if while connected I look for the image from the remote DB, as a matter of fact while connected I will use the Framework EntityManger “normally”

So now disconnected, I have a field rep that will be going to a customer site, the site may not grant / have internet access (to the remote DB) so before he leaves he will perform an operation that will hydrate all the related “customer” specific information related to the project. No big deal I can use the cache file for this perfect. The only caveat is if the tek needs to lookup a part from a catalog, this catalog is huge both general data as well as pictures, and the kicker is that the tek needs the ability to add a new entry including a picture as well, this new entry will need to be pushed back to the remote DB and all the other tech need to syc up when it is available.

So maybe the catalog DB piece is different DB than the main app. That way I can use a separate EntityCache and EntityStateFile for the piece. The workflow for this piece would be: when connected use the remote DB, when disconnected using a separate EntityManager look in the local DB and use the EntityManager as designed and save the CacheFile on exit. Then when connected I could restore the cache file then save data to the remote DB.

Will this work?

 I think I will.

Is this the same solution that is in the doc? Maybe so. Just took me awhile to get there.

Thanks

Dave